Video on demand (VOD) is not a new concept. It has been implemented in small scale, closed systems for a number of years (as used in this application, closed system means additional users cannot be added after the system has been installed). In particular, VOD has been used in hotels, on airplanes, and in other, similar closed systems. Recently, larger scale television services have begun implementing VOD services on larger, open television systems (as used in this application, open system means additional users can be added after the system has been installed). U.S. Pat. No. 5,850,218, issued Dec. 15, 1998, to Lajoie et al. for “Inter-Active Program Guide with Default Selection Control,” incorporated herein by reference, describes one possible television network capable of supplying VOD services. A television system in Orlando, Fla. has implemented a VOD system similar to the one described in U.S. Pat. No. 5,850,218.
As is evident in the above examples, conventional wisdom for implementing VOD services in television systems contemplated a centralized service. The centralized VOD service model placed (or housed) the computers and media servers necessary to implement and control the VOD services in the central or headend location. Customers would access the servers through their set-top boxes. The set-top boxes would communicate with the media server at the headend to establish a video stream for the requested video. On a smaller scale with a limited number of users, such as the hotel system mentioned above, the centralized service works fairly well. Centralized VOD services work in the smaller environment partially due to the predictable and limited number of video streams that are necessary to satisfy expected peak demand for VOD services. Additionally, the content of the smaller centralized server is typically limited to “new release” types of movies.
However, implementing the centralized system on a large and potentially ever increasing open network system is prohibitively expensive because of the large size of the content transport network required to support the number of interactive video sessions. For example, in a network with 500,000 customers and assuming 10% (50,000) are simultaneous users, 20,000 Mbps must be available from the headend to the hubs to support the expected peak demand of the VOD services. Increasing demand causes a corresponding increase in the number of video streams that the centralized server needs to provide. Furthermore, in the typical wide area television network, the number of subscribers continually increases. As the number of subscribers increases or the demand expands, it becomes increasingly difficult to provide the server streaming capacity and the bandwidth and switching capability of the interactive transport network to accommodate the increase in users. It would therefore be beneficial to develop a content delivery system having VOD services that was not reliant solely on the centralized media source.
By distributing servers to hubs (closer to the customers), the size of the fiber transport network can be reduced, which significantly reduces the costs associated with implementing a large-scale interactive VOD service. A purely distributed system, however, has some of its own disadvantages. As all servers must provide all and any content that is promoted as available on the system, each server must provide a large storage capacity that must include low-demand library titles. Such content numbering in hundreds or thousands of titles consumes the vast majority of storage capacity, yet it generates very limited revenues. It can be documented that 90% of such content generates only 10% of the earnings, while the current 10% of the high-demand content generates 90% of the total revenues. In addition, a purely distributed system cannot realistically support interactive VOD functionality for real time events that would provide personal video recorder (PVR) experience for live events. Therefore, it is desirous to develop and engineer a hybrid central/distributed VOD system with a tiered content structure that would provide a vastly superior technical and economical performance. It segregates the content according to its demand and makes each of the high and low demand content available in a most economic and reliable fashion to the users. It would further enable fully interactive real time event VOD services, providing PVR functionality. The hybrid central/distributed VOD system with a tiered content structure is applicable to interactive content delivery systems used in cable systems, as well as geographically larger delivery systems such as internet, satellite or regional DSL service systems. For example, as set top boxes and satellite receivers with record and replay capability are becoming available, the term “hub server” as it relates to high demand content storage may be supplemented by any of these home devices. One may assume high demand content to be pushed to homes (as it is now file transferred to the hub servers) while the low demand content is ordered on an individual basis (as it is now from central server).